How many SNPs are in the human genome?
Per person, common in the population, and catalogued in dbSNP
Matic Broz, PhD7 min
A typical person carries about 4 million to 5 million SNPs, positions where their DNA has a different single base from the human reference genome. Across everyone, about 10 million SNP sites are common, meaning the less frequent base is found in at least 1% of people. The largest figure belongs to NCBI's dbSNP database, whose Build 157 lists 1.17 billion reference variant records because it collects rare as well as common variants from many sequencing projects.
These three numbers measure different things: one genome, the shared variation of a population, and a catalogue of everything reported so far. Mixing them up is why answers to this question range from a few million to more than a billion.
What is a SNP?
A single-nucleotide polymorphism, or SNP (pronounced "snip"), is a position in the genome where people differ by one base. DNA is written in four bases, A, C, G and T, and at a SNP some people carry, for example, a C where others carry a T. The human genome has about 3.1 billion bases per chromosome set, so even millions of SNPs touch only a small fraction of it.
Strictly, a SNP is a single-base variant common enough to count as a polymorphism. The National Human Genome Research Institute reserves the name for single-nucleotide variants (SNVs) present in at least 1% of the human population, so a rarer one-base change is an SNV but not a SNP.[1] Many researchers and databases use "SNP" loosely for any single-base variant, which is one reason published counts disagree. Frequency also depends on who is sampled, since a variant common in one ancestry group can be rare in another.
A SNP is also not the same as a mutation. "Mutation" names the event that changed the DNA, or the resulting change, and says nothing about how common the change is or whether it affects health. The Human Genome Variation Society therefore recommends neutral terms such as "variant" when the origin and effect of a change are not established.[2] Most SNPs have no known effect. For one that falls in regulatory DNA, AlphaGenome can predict effects on molecular signals, though a prediction alone does not show that a variant causes disease.
How many SNPs does one person have?
A typical genome differs from the reference at about 4 million single bases. When the 1000 Genomes Project resequenced 3,202 genomes at 30-fold coverage, the average sample carried 4,080,992 SNVs, along with 451,277 small insertions or deletions.[3]
The count depends on ancestry. Genomes from African populations averaged 4.65 million SNVs, about 0.8 million more than European or East Asian genomes.[3] The 1000 Genomes Consortium noted the same pattern in its earlier data and attributed it to the out-of-Africa model of human origins: populations in Africa have the longest history and so the greatest genetic diversity.[4]
| Ancestry group | Average SNVs per genome |
|---|---|
| African | 4,653,521 |
| Admixed American | 3,911,413 |
| South Asian | 3,896,324 |
| East Asian | 3,822,328 |
| European | 3,818,951 |
| All samples | 4,080,992 |
Other estimates agree on the scale. The original 1000 Genomes analysis of 2,504 people found that a typical genome differed from the reference at 4.1 million to 5.0 million sites, more than 99.9% of them SNPs or short insertions and deletions.[4] The National Human Genome Research Institute rounds the per-person figure to about 5 million SNVs and 600,000 small insertions or deletions.[1] Counts shift with the reference assembly, sequencing depth and variant-calling rules, so differences of a few hundred thousand between studies are expected rather than contradictory.
Most of these single-base differences are SNPs in the strict sense as well. Only 40,000 to 200,000 variants in a typical genome, 1% to 4% of the total, had a frequency below 0.5% in the 1000 Genomes data; the rest are shared with many other people.[4]
Each person carries two copies of most chromosomes, one from each parent, and the per-person count includes two kinds of site. At a heterozygous site the two copies carry different bases; at a homozygous site both carry the same non-reference base. In the Genome in a Bottle reference sample HG002, a 2025 analysis found 3,956,307 SNVs with a heterozygous-to-homozygous ratio of 1.56, which works out to about 2.41 million heterozygous and 1.55 million homozygous sites.[5] Finding these sites requires a variant caller working from sequencing reads; aligning two assembled genomes with MUMmer4 shows the same kinds of differences without saying how common they are in the population.
How many common SNPs are there?
About 10 million SNP sites are common in the human population. The International HapMap Project used this estimate for sites where both bases occur at a frequency of at least 1%, roughly one every 300 bases, and noted that these common SNPs account for about 90% of human genetic variation.[6]
Large-scale sequencing has since tested that figure. Among 2,504 genomes, the 1000 Genomes Project found about 8 million autosomal variants with a frequency above 5% and about 12 million between 0.5% and 5%.[4] The number above 1% must therefore lie between 8 million and 20 million, consistent with the HapMap estimate. The exact count depends on the cutoff and on whether a variant must be common worldwide or only in one population; a 5% threshold, for instance, gives a figure closer to 8 million.
A single person carries fewer differences than there are common SNPs because at many common sites a person simply has the reference base on both copies. Any two copies of the human genome differ at about one base in 1,000, a small fraction of the positions where humans vary.[6]
Why does dbSNP contain more than a billion records?
dbSNP Build 157 contains 1,172,689,405 live Reference SNP records and 4,849,775,973 submitted records. NCBI announced the build in March 2025 with updated data from 1000 Genomes, TOPMed, gnomAD and its own ALFA allele-frequency project.[7] It was still the latest release on NCBI's download server when we checked on October 23, 2026.[8]
A submitted record, identified by an ss number, represents one study's submission. NCBI merges submissions that describe the same mapped variant into a non-redundant Reference SNP cluster with an rs number, which is why there are about four times as many submitted records as reference records.
Despite its name, dbSNP is not limited to SNPs or to common variants. Its reference clusters also represent multi-nucleotide variants, insertions and deletions, and it includes rare as well as common variants, so the Build 157 total should not be read as 1.17 billion common SNPs.[9]
The catalogue keeps growing because nearly every newly sequenced genome adds variants seen in only a few people. The 1000 Genomes Project found 84.7 million SNPs among 2,504 genomes in 2015, and its high-coverage resequencing of 3,202 genomes found 111 million SNVs.[4][3] The per-person count, by contrast, stays near 4 million to 5 million however many genomes are added.
Taken together, the numbers form a ladder. One genome differs from the reference at 4 million to 5 million single bases, almost all of them shared with other people. About 10 million sites are common across humanity. Rare variation is far larger: more than 100 million single-base variant sites appear in a few thousand genomes, and more than a billion variant records have accumulated in dbSNP. When a source quotes a number of SNPs, the first thing to check is which of these scopes it describes.
